At the recently concluded 2026 electronica China, embodied intelligence and core robot components (precision joints, dexterous hands, servo control, etc.) became the focal point of the event. Multiple domestic MCU (Microcontroller Unit) manufacturers showcased their latest real-time control chips, dedicated motor drivers, and high-precision sensing solutions designed for embodied intelligence.
By reviewing the technical solutions and product portfolios showcased by various MCU manufacturers at the exhibition, we conducted an in-depth comparison of the solutions from leading MCU players including GigaDevice, Nation Technologies, HPMicro, Artery, Cmsemicon, and Geehy.
Core Insights into Domestic MCUs in the Field of Embodied Intelligence:
Domestic MCUs are embracing explosive growth opportunities in embodied intelligence. The chip BOM (Bill of Materials) value for a single humanoid robot can reach $500. Domestic manufacturers have significantly exceeded market expectations in their penetration rate for key components such as servo drives and dexterous hand control boards. They have successfully entered the supply chains of leading robot OEMs by leveraging rapid customization responses and ultimate cost-effectiveness.
Computing performance is showing a multipolar development trend. Nation Technologies' N32H7 series, featuring a dual-core M7+M4 architecture with a clock speed of 700MHz, ranks at the top domestic level. GigaDevice has built a diverse core matrix of M7/M33/RISC-V to maintain ecosystem leadership. Artery has pushed the M4 clock speed to 288MHz to explore extreme performance. All manufacturers have integrated hardware trigonometric function accelerators and SVPWM (Space Vector Pulse Width Modulation) generators to significantly improve the execution efficiency of FOC (Field-Oriented Control) algorithms.
The EtherCAT real-time bus has become a critical threshold for the high-end control layer. GigaDevice's GD32H75E and Nation Technologies' N32H78x both deeply integrate EtherCAT slave controllers to achieve microsecond-level multi-joint synchronization. CAN-FD has become standard for high-end MCUs across all manufacturers to meet the high-frequency distributed data interaction needs of robots.
Differentiated packaging precisely addresses application pain points. Artery introduced the 7×7mm BGA100 micro-package AT32F435/437, directly targeting the space constraints of dexterous hands. Cmsemicon helps robot vacuum cleaner manufacturers reduce supply chain BOM costs by approximately 20% through a highly integrated SoC (System on Chip) solution that integrates LDO (Low Dropout) and N+N drivers.
The strategic layouts of the six major manufacturers show distinct differentiation: GigaDevice ranks first domestically relying on its "MCU+Flash" platform-level ecosystem and complete software stack; HPMicro focuses on RISC-V high computing power and GPU graphics acceleration to create a differentiated route; Nation Technologies integrates financial-grade security features into the Industry 4.0 foundation; Artery, relying on ultra-micro packaging, has been mass-adopted by leading robot manufacturers such as Unitree Robotics; Cmsemicon lowers the customer development threshold through extremely integrated SoCs and graphical debugging tools; Geehy adopts the cutting-edge Cortex-M52 architecture, with its Helium technology significantly enhancing edge AI and signal conditioning capabilities.
Domestic MCU security certifications have achieved phased breakthroughs, but high-end fields still face pressure from international giants. Multiple products from GigaDevice have passed IEC 61508 SIL2/SIL3 certifications, and Nation Technologies has obtained the ISO 26262 ASIL-D automotive-grade certification. However, in high-safety fields such as powertrains and chassis, the market is still dominated by giants like Infineon AURIX, STMicroelectronics STM32, and Texas Instruments C2000.
Competitive Landscape and Selection Matrix of Domestic MCUs at the Embodied Intelligence Booths
1. GigaDevice: The "Full-Stack Integrator" with Multi-Dimensional Layout
Focusing on joint drive, force sensing, and power management in embodied intelligence, GigaDevice showcased comprehensive robot application solutions based on its high-performance GD32 MCUs and analog chips.
Displayed Solutions and Core Products:
Six-Axis Robotic Arm Solution Based on GD32H75E: Composed of joint drivers and a host main controller. The GD32H75E handles the underlying servo control of joint motors, equipped with 17/18-bit dual absolute encoders, supporting CoE (CANopen over EtherCAT) communication and the CiA 402 protocol; the host main controller is responsible for trajectory planning and collision detection.
GD30AD3642 High-Precision ADC Six-Dimensional Force Detection Solution: A signal acquisition solution specifically designed for robot force/tactile sensing.
GD30DC1901 100V/1A High-Voltage Synchronous Buck Power Solution.
Pros and Cons in Technology and Application:
Advantages: Extremely strong full-stack ecosystem capabilities. It possesses one-stop supply capabilities for "MCU + Memory + Analog/Power + Sensors", which facilitates robot manufacturers in optimizing overall BOM costs. Its GD32H75E (Cortex-M7 core, 600MHz) features ultra-high computing power and rich analog interfaces. The support for dual absolute encoders and the six-dimensional force detection solution make the closed-loop of high-precision control and sensing tighter.
Disadvantages: The traditional core architecture lacks hardware-level dedicated coprocessor acceleration for specific robot kinematics algorithms (such as trigonometric functions and coordinate transformations). Moreover, compared to some RISC-V dual-core or drive-control integrated solutions, it does not have an absolute advantage in PCB (Printed Circuit Board) layout space for ultra-compact finger joints (dexterous hands).
2. Nation Technologies: The "Pioneer of Distributed Control" with "Division of Labor Between 'Big Brain' and 'Small Brain'"
Nation Technologies regards its integrated foundation of "MCU + Security Chip + Wireless RF + BMS" as its strategic core. At the exhibition, it focused on showcasing distributed control solutions serving dexterous hands and humanoid robot joints.
· Displayed Solutions and Core Products:
o "1+4" Joint/Dexterous Hand Architecture Based on N32H7 and N32H4 Series: Uses 1 N32H7 (China's first Cortex-M7+M4 dual-core heterogeneous MCU, 600MHz) and N32H4 chips, responsible for local closed-loop control and FOC algorithm execution of a single joint.
o EtherCAT Multi-Axis Linkage Servo Control Solution: Showcased based on N32H785EC.
o Application Cases: Mass-deployed in Honor's "Yuanqizai" humanoid robot (which completed a 21km half-marathon) and Dobot's "Rover X1" home intelligent agent.
· Pros and Cons in Technology and Application:
o Advantages: Hardware-level acceleration and high security. It has a built-in CORDIC (Coordinate Rotation Digital Computer) coprocessor that specifically handles coordinate rotation and trigonometric calculations in kinematics via hardware, significantly reducing CPU load. The 100ps-class ultra-high-resolution timer ensures the stability of millisecond-level balance control. In addition, it integrates a hardware-level security chip, meeting the requirements for robot privacy and security in home and specific sensitive fields.
o Disadvantages: The "1+4" multi-chip distributed architecture has higher multi-core/multi-device programming complexity in software and hardware development, requiring developers to have strong system-level debugging capabilities.
3. HPMicro: The "RISC-V Computing Power Monster" Breaking Space Bottlenecks
HPMicro focuses on high-performance, high-bandwidth robot joints and motor control, committed to breaking the zero-sum game between "computing power" and "volume" through its independent RISC-V architecture.
· Displayed Solutions and Core Products:
o HPM6E8Y Series Dual-Core RISC-V Processor: Clock speeds up to 600/800MHz, it is China's first MCU to obtain official German EtherCAT Slave Controller (ESC) licensing. It integrates 2 high-speed Ethernet PHYs (Physical Layers), a Gigabit TSN (Time-Sensitive Networking) switch, 32-channel 100ps resolution PWM, and a 16-bit ADC. A single chip supports up to 4-axis motor control.
o HPM53M1 Joint-Dedicated "Drive-Control Integrated" MCU: Clock speed of 480MHz, integrating 4x CAN-FD, two 16-bit ADCs, and high-performance three-phase independent half-bridge pre-drivers.
· Pros and Cons in Technology and Application:
o Advantages: Performance and integration reach the industry's top tier. The HPM6E8Y directly integrates the EtherCAT ESC and physical layer PHY, eliminating the need for additional communication chips in robot joint wiring and significantly reducing PCB area. The HPM53M1 achieves "drive-control integration" with built-in pre-drivers, greatly simplifying hardware design in extremely small spaces such as joints and dexterous hands. The HPM_MCLV2 motor library supports hardware current loops, DQ-axis decoupling, and PLL (Phase-Locked Loop).
o Disadvantages: Although it adopts the RISC-V architecture with strong performance, for traditional industrial control developers accustomed to the traditional Arm development environment (such as Keil/IAR), there is still a certain learning curve for software ecosystem migration and toolchain familiarization.
4. Artery: The "Cost-Effective Dexterous Hand" with "Small Package and Multi-Axis Control"
At the exhibition, targeting high-DOF dexterous hands and compact robot joints, Artery showcased its deep layout in small packaging and motor control resources.
· Displayed Solutions and Core Products:
o AT32F435/437 Series Adds BGA100 (7×7 mm) Small-Size Package: Clock speed up to 288MHz (Cortex-M4F), specifically designed for extremely constrained spaces such as robot finger joints.
o Motor-Dedicated AT32M412/M416 Series: Clock speed of 180MHz, built-in high-performance Op-Amps (Operational Amplifiers) to directly process sensor analog signals.
o Application Solutions: A single MCU, through built-in Advanced Timers and high-speed ADCs, can support up to 3 sets of motor control simultaneously, achieving multi-axis synchronization.
· Pros and Cons in Technology and Application:
o Advantages: Ultimate space utilization and cost-effectiveness. The 7×7 mm BGA100 package is a powerful tool to solve the PCB layout pain points of dexterous finger joints; a single chip controlling three motors greatly reduces the number of control nodes and the overall BOM cost. It is equipped with user-friendly graphical monitoring and debugging tools (AT32 Motor Monitor), lowering the development threshold.
o Disadvantages: The communication interfaces are mainly based on CAN 2.0B and UART, without native integration of high-speed Ethernet (such as EtherCAT/TSN) or CAN-FD. In complex systems requiring hundreds of axes synchronization with extremely high data throughput, network bandwidth may become a bottleneck.
5. Cmsemicon: The "Edge Nerve Endings" with "Ultra-Low Power Consumption"
Cmsemicon focuses on energy efficiency upgrades and device miniaturization, showcasing dedicated low-power chips for lightweight joint control and peripheral actuators.
· Displayed Solutions and Core Products:
o BAT32G139 Series Ultra-Low Power MCU: Adopting the Cortex-M0+ core with a clock speed of 64MHz, it integrates two CAN channels, comparators, PGA (Programmable Gain Amplifier), and 17 16-bit timers with motor-dedicated PWM.
o CMS32M66xx Series Highly Integrated Motor Control Chips.
· Pros and Cons in Technology and Application:
o Advantages: Ultimate power consumption control and analog integration. The operating power consumption is only 120uA/MHz, and the power consumption in deep sleep is as low as 0.8uA. It is extremely suitable for battery-powered mobile robots (AGV (Automated Guided Vehicle) / AMR (Autonomous Mobile Robot)), companion robots, or edge sensor fusion nodes that require long-term standby. Highly integrated analog peripherals reduce the cost of external components.
o Disadvantages: The computing performance is relatively weak (64MHz M0+), unable to support complex real-time inverse kinematics calculations for robots or high-bandwidth multi-axis and multi-sensor fusion. Its positioning leans more towards simple actuators at the robot's edge (such as electric grippers, cooling fan control, or sensor preprocessing).
6. Geehy: The "Rising Star" with "Hardware-Based Algorithms"
Targeting the full chain of "sensing-computing-controlling" in embodied intelligence, Geehy showcased its latest generation of MCUs with built-in AI acceleration and self-developed instructions, demonstrating strong innovation capabilities.
· Displayed Solutions and Core Products:
o G32R501 Real-Time Control MCU/DSP (Digital Signal Processor): Based on the Cortex-M52 dual-core architecture (250MHz), it integrates the Arm Helium edge AI acceleration unit, significantly improving DSP and machine learning efficiency.
o G32R430 Encoder-Dedicated MCU: Integrates self-developed ATAN (arctangent) electrical angle calculation extension instructions, with an electrical angle delay of less than 1μs and an angle accuracy better than 0.0001°.
· Pros and Cons in Technology and Application:
o Advantages: Cutting-edge architecture and dedicated computing power optimization. The G32R501 is one of the few domestic chips adopting the Cortex-M52, and its Helium technology significantly enhances edge AI and signal conditioning capabilities; the G32R430 solves the delay problem of high-frequency, high-precision reading of joint encoders through dedicated self-developed ATAN instructions in a hardware-hardened manner.
o Disadvantages: Cortex-M52 and Helium are relatively new technology architectures. Their supporting third-party development ecosystems and accumulation of mature cases are still in the early stages, requiring a certain amount of time for technology popularization.
Overall, domestic MCU manufacturers have completely broken away from the red ocean competition of simple "Pin-to-Pin" consumer-grade domestic substitution in the past. On the golden track of embodied intelligence, they are launching a strong assault against the high-end barriers of "high computing power, high integration (drive-control integration), high-bandwidth communication (built-in EtherCAT), and hardware-based algorithms (dedicated coprocessors/custom instructions)."